ELHS Teachers Adapting to AI’s Effects on Students’ Learning
As technology continues to play a greater role in education, local high school teachers are adapting to the use of ChatGPT in the classroom. This AI language model, developed by OpenAI, has the ability to generate human-like text based on its training on vast amounts of data. Teachers in the area are exploring how ChatGPT can be leveraged to improve student engagement, enhance communication skills, and streamline administrative tasks.
You’ve probably read a paragraph like the one above and, whether you knew it or not, it may have been generated by ChatGPT — as this one was. Like many of the answers ChatGPT provides, when I asked it to “write the lede for an article for a local non-profit news service about how chatgpt will be handled by local high school teachers,” it provided a very intelligent sounding, but only partially correct, answer.
ChatGPT is an artificial intelligence-driven service from the company OpenAI released late last year. OpenAI was founded by Elon Musk and other notable Silicon Valley entrepreneurs for the purpose of promoting artificial intelligence that, according to its mission statement, is “to ensure that artificial intelligence benefits all of humanity.” (Musk is no longer involved). As part of its mission, it restricts the types of questions ChatGPT will answer to mitigate negative outcomes like violence and discrimination.
Yet the very existence of ChatGPT, and its ability to generate coherent text on many topics, inherently presents a problem to teachers who evaluate students based on written performance. Many large school districts, like New York City’s, quickly banned the use of ChatGPT and blocked access to the service on district devices and networks.
According to Bridge Michigan, an English teacher at Okemos High School already suspects multiple students have submitted papers generated by ChatGPT.
ELi spoke with teachers at East Lansing High School (ELHS) about ChatGPT to learn if it has impacted academic dishonesty and to learn whether they are choosing to use it in the classroom.
Teachers know that cheating is not a new problem in education, but some say it has proliferated in recent years, driven by technology and the move to remote learning during the pandemic. And ChatGPT takes the ability to another level.
ELHS math teacher Jeff Burgess provided this example: Math students can utilize apps that solve equations by taking a picture with their phone; ChatGPT allows students to give in-depth explanations with sound mathematical reasoning. Previous apps could get the “right” answer but didn’t help students say how they got it. ChatGPT gets to the “why.”
Cheating could often be caught and was riskier when all teachers had to do was Google an answer or piece of writing they suspected was plagiarized. Because ChatGPT generates unique text, ELHS English and Psychology teacher Jade Bennett believes student cheating will be harder to detect, and more time-consuming.
OpenAI recently published a tool to help teachers detect cheating. But the company does not recommend it be used as “the sole piece of evidence, when deciding whether a document was generated with AI.” The company warns the tool is not accurate enough, saying, the “model is trained on…a variety of sources, which may not be representative of all kinds of human-written text.”
ELHS Social Studies teacher Mark Pontoni has found he can ask ChatGPT directly if it wrote an answer given to him by a student. That, combined with comparing the answer to a student’s other work, has been a reliable but time-consuming way to detect cheating.
Teachers at ELHS are adapting their methods to prevent cheating.
Bennett and Pontoni have stopped assigning written assignments as homework, requiring it to be done in class. This, according to Pontoni, “has negative long-term effects on education because I can’t do everything in class that I used to do.”
While not all teachers have evidence of it in their classes, Pontoni’s method has uncovered multiple assignments where ChatGPT was used, including one case in which a student turned in an assignment with the “Regenerate Response” link from ChatGPT following the paragraph.
Students do not necessarily understand this is cheating, and school administration and teachers acknowledge they need to do more to teach students what is and is not acceptable.
ELHS has a plagiarism policy in its policy handbook (on page 45, item 8, under the Prohibited Student Conduct and Student Discipline). According to Interim Principal Ashley Schwarzbek, this policy, in spirit, covers the usage of AI. But she acknowledges ELHS will look to add “additional language that specifically addresses AI when we make handbook updates this summer.”
Enforcement of academic dishonesty is handled both by teachers in the classroom and by the administration. Pontoni believes adjustments to the policy, along with a demonstrated commitment by the administration, will lead teachers to be more confident that they will have the support needed to tackle the challenge.
ChatGPT has the potential to enhance or harm students’ education.
Pontoni, who has shown students how they can use AI tools for background information, views ChatGPT “like any other piece of technology that students have used to lighten their workload.” He encourages other teachers to both emphasize that it must not be attributed as students’ original work and to embrace how it could improve education.
Meanwhile, ELHS’ administration is trying to better understand the impact of AI on education. It sent one teacher to a professional development opportunity on ChatGPT recently, who will then share back what they’ve learned with all staff.